Qdrant

Qdrant is an open-source, high-performance vector database and similarity search engine written in Rust, designed for production-grade AI applications that require precise control over retrieval quality, filtering, and scaling. Its primary differentiators over simpler vector stores are the filtering capability and payload indexing: you can combine vector similarity search with exact field constraints at query time without post-processing the results, enabling queries like 'find the most similar document to this embedding, from the subset authored by this user, created in the last 30 days' in a single efficient operation. Qdrant supports multiple distance metrics (cosine, Euclidean, dot product) and multiple vector types including sparse vectors for hybrid dense-sparse retrieval pipelines. The Quantization feature reduces memory usage by 4–32x with minimal accuracy loss, enabling large vector collections to fit on cost-effective hardware.

Sparse vector support enables BM25-like keyword retrieval alongside dense semantic search in the same engine. Deployment options include self-hosted on Kubernetes or Docker, or Qdrant Cloud for managed operation. The open-source license has no commercial restrictions. Qdrant is commonly chosen for production RAG systems, recommendation engines, and semantic search applications where the filtering precision, Rust-level performance, and operational predictability outweigh the setup simplicity of alternatives like Chroma.